Construction and Application of a Prognostic Model for Lung Cancer Brain Metastasis Based on Plasma miRNA Combinatorial Biomarkers

Abstract

The present invention discloses the construction and application of a prognostic model for lung cancer brain metastasis based on a plasma miRNA combination biomarker. The present invention constructs a prognostic model for lung cancer brain metastasis based on a plasma miRNA combination biomarker, and the plasma miRNA combination biomarker includes miR-31-5p, miR-219-5p, miR-1-3p, miR-130a-5p, miR-452-3p, miR-598-3p, miR-9-5p, miR-1246, miR-196a-5p and miR-210-3p. The prognostic model obtains a miRNA expression index according to the expression levels of each plasma miRNA, and obtains a predicted survival rate as a prognostic prediction result according to the miRNA expression index and clinical information; this prognostic model is used for prognostic prediction of lung cancer brain metastasis. The prediction model and prognostic prediction system in the present invention both have high sensitivity and accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of medical artificial intelligence, and particularly relates to the construction and application of a prognostic model for lung cancer brain metastasis based on plasma miRNA combined markers. Background Art

[0002] Among patients with solid tumor brain metastases, lung cancer brain metastases rank first among brain metastases with a metastasis incidence rate as high as 40%, and the prognosis is extremely poor. Lung cancer brain metastases can be divided into parenchymal metastases and meningeal metastases, and their common clinical manifestations are mainly increased intracranial pressure, specific focal symptoms and signs. At present, the diagnosis of brain metastases mainly relies on imaging techniques such as head magnetic resonance imaging. However, since the resolution of head magnetic resonance imaging is in the millimeter range, while the size of tumor cells is in the micron range, this will lead to delayed diagnosis and treatment of tumors clinically.

[0003] At present, lung cancer brain metastasis patients can choose comprehensive treatment techniques such as surgery, radiotherapy and chemotherapy, which prolong the survival period of patients. However, the median survival period of related patients is still short, and it is about 12 to 15 months after treatment. Based on the characteristics of lung cancer brain metastases with insidious onset, rapid progression, poor prognosis and short natural survival period, there is an urgent need to screen new biomarkers and construct a prediction model for the diagnosis and prognosis of lung cancer brain metastases.

[0004] Serum tumor markers related to lung cancer include carcinoembryonic antigen, cytokeratin fragment 19 and squamous cell carcinoma antigen, etc. These serum biomarkers can assist in monitoring the disease changes and drug treatment effects of lung cancer patients. However, these indicators are easily affected by the microenvironment in the patient's body and there is a certain false positive rate. In addition, although the development of radiomics can provide more information for lung cancer brain metastasis patients, including features such as the number, volume and location of lesions in imaging data, obtaining this information has a large computational workload and requires professional radiologists to set parameters and perform scans. In summary, there is currently no simple, efficient and excellent prognostic prediction system for lung cancer brain metastasis to accurately evaluate the prognosis of lung cancer brain metastasis patients.

[0005] In recent years, the research on plasma or serum microRNA (miRNA) as a biomarker for tumor prognosis has received continuous attention. miRNA is about 20 - 25bp in length and affects the expression of target genes at the post-transcriptional level, widely participating in physiological processes such as the proliferation, differentiation, and apoptosis of tumor cells. In the field of lung cancer brain metastasis, many studies have shown that miRNA is involved in the invasion and migration of lung cancer, and miRNA as a new biomarker can guide the prognosis of patients with highly invasive lung cancer. The existing patent publication number CN108728543A discloses a miRNA combination for detecting lung cancer brain metastasis and related kits. This method detects 3 miRNA molecular markers in cerebrospinal fluid, and the discrimination of the risk of lung cancer brain metastasis based on this has a certain effect, but it cannot be applied to the precise prognosis of patients with lung cancer brain metastasis.

[0006] Therefore, there is currently no potential simple, efficient, and excellent predictive performance prognostic prediction system and method for lung cancer brain metastasis considering miRNA. Summary of the Invention

[0007] Aiming at the problem that the existing prediction systems and methods cannot be applied to the precise prognosis of patients with lung cancer brain metastasis, the present invention provides a construction and application of a prognostic model for lung cancer brain metastasis based on plasma miRNA combination markers. The prognostic model provided by the present invention is constructed based on plasma miRNA combination markers and clinical information and is used for the prognostic prediction of lung cancer brain metastasis.

[0008] The present invention adopts the following technical solutions:

[0009] I. A plasma miRNA combination marker for prognostic prediction of lung cancer brain metastasis

[0010] The plasma miRNA combination marker mainly consists of miR - 31 - 5p, miR - 219 - 5p, miR - 1 - 3p, miR - 130a - 5p, miR - 452 - 3p, miR - 598 - 3p, miR - 9 - 5p, miR - 1246, miR - 196a - 5p, and miR - 210 - 3p.

[0011] The plasma miRNA is the miRNA in an in vitro biological plasma sample.

[0012] In the plasma miRNA combination marker, miR - 31 - 5p, miR - 219 - 5p, miR - 1 - 3p, miR - 130a - 5p, and miR - 452 - 3p are highly expressed in in vitro biological plasma samples of lung cancer brain metastasis. miR - 598 - 3p, miR - 9 - 5p, miR - 1246, miR - 196a - 5p, and miR - 210 - 3p are low - expressed in in vitro biological plasma samples of lung cancer brain metastasis.

[0013] II. Application of the above plasma miRNA combination marker

[0014] The plasma miRNA combination marker is used for preparing a prognostic prediction drug or a prognostic prediction system for brain metastasis of lung cancer.

[0015] The prognostic prediction drug includes primers, probes or fluorescent reagents capable of specifically detecting all plasma miRNA markers.

[0016] The prognostic prediction system includes a Cox regression prognostic prediction model.

[0017] III. Construction method of a prognostic model for brain metastasis of lung cancer based on a plasma miRNA combination marker

[0018] The construction method includes the following steps:

[0019] 1) Obtain sample data from at least one bioinformatics database, perform differential expression analysis on the sample data obtained from each bioinformatics database to obtain a differential expression miRNA combination corresponding to each bioinformatics database; extract the intersection of the differential expression miRNA combinations corresponding to all bioinformatics databases to obtain a common differential expression miRNA combination of all bioinformatics databases, and screen out a significant common differential expression miRNA combination from the common differential expression miRNA combination according to a preset threshold as a candidate miRNA combination;

[0020] Specifically, from the common differential expression miRNA combination, screen out miRNAs with a fold change greater than a preset fold change threshold and a statistical significance less than a preset statistical significance threshold as significant common differential expression miRNAs, that is, candidate miRNAs.

[0021] 2) Construct the expression levels of each candidate miRNA in pre-collected normal ex vivo biological plasma samples and lung cancer brain metastasis ex vivo biological plasma samples into a sampling data set, perform differential expression analysis on the sampling data set to obtain multiple differentially expressed miRNAs, and form a plasma miRNA combination, and the plasma miRNA combination is used as a combination marker for prognostic prediction of brain metastasis of lung cancer; use the Logistic regression method to construct a prediction model of miRNA expression index according to the plasma miRNA combination and the expression levels of each plasma miRNA therein;

[0022] In step 2), pairwise correlation analysis is performed on the sampling data set and collinear miRNAs are removed, and then differential expression analysis is performed;

[0023] In step 2), each plasma miRNA is miR-31-5p, miR-219-5p, miR-1-3p, miR-130a-5p, miR-452-3p, miR-598-3p, miR-9-5p, miR-1246, miR-196a-5p, and miR-210-3p;

[0024] In step 2), the prediction model of miRNA expression index is specifically:

[0025] g= 7.97 - 4.45 m 1 -2.85 m 2 +1.82 m 3 -3.38 m 4 -3.18 m 5 +1.13 m 6 +6.58 m 7 +4.67 m 8 +1.52 m 9 -7.73 m 10

[0026] In the formula, m 1 、 m 2 、 m 3 、 m 4 、 m 5 、 m 6 、 m 7 、 m 8 、 m 9 、 m 10 respectively represent the expression levels of miR-31-5p, miR-219-5p, miR-1-3p, miR-130a-5p, miR-452-3p, miR-598-3p, miR-9-5p, miR-1246, miR-196a-5p, and miR-210-3p.

[0027] 3) Perform univariate Cox regression analysis and multivariate Cox regression analysis on the clinical information corresponding to all in vitro biological plasma samples of lung cancer brain metastases in sequence, obtain multiple clinical variables, and form a clinical variable combination; use the Cox regression analysis method to construct a Cox regression prognosis prediction model based on the clinical variable combination and miRNA expression index, and use the K-fold method to optimize the Cox regression prognosis prediction model to obtain the said prognosis model.

[0028] In the said step 3), the clinical variable combination includes age, KPS, and GPA.

[0029] In the said step 3), the prognosis model is specifically:

[0030] h(t)=h 0 (t) exp( PI )

[0031] PI =1.43 X 1 -0.28 X 2 -22.27 X 3 +58.72 X 4

[0032] In the formula, h(t) represents the risk function, h 0 (t) represents the baseline risk function, X 1 represents age, X 2 represents KPS, X 3 represents GPA, X 4 represents the miRNA expression index, t represents the predicted survival time, PI represents the prognosis index.

[0033] IV. A prognosis prediction method for lung cancer brain metastases based on the above plasma miRNA combined marker

[0034] The said prognosis prediction method includes the following steps: Obtain the expression levels of each plasma miRNA in the in vitro biological plasma sample and the clinical information corresponding to the in vitro biological plasma sample, input all the expression levels of plasma miRNA and the clinical information into the prognosis model. The prognosis model obtains the miRNA expression index based on all the expression levels of plasma miRNA, and then obtains the predicted survival rate according to the miRNA expression index and the clinical information and outputs it as the prognosis prediction result.

[0035] The predicted survival rate is obtained by the following formula:

[0036] S(t) = exp(- ∫ t 0 h 0 (u)du × exp( PI ))

[0037] In the formula, S(t) represents the predicted survival time t of the predicted survival rate, t represents the predicted survival time, ∫ t 0 h 0 (u)du represents the predicted survival time t within the cumulative risk, PI represents the prognostic index.

[0038] Among them, the prognostic index PI is obtained by the following formula based on the miRNA expression index and clinical information. The clinical variables in the clinical information include age, KPS, and GPA:

[0039] PI = 1.43 X 1 - 0.28 X 2 - 22.27 X 3 + 58.72 X 4

[0040] In the formula, X 1 represents age, X 2 represents KPS, X 3 represents GPA, X 4 represents the miRNA expression index.

[0041] Among them, the miRNA expression index X 4 is obtained by the following formula:

[0042] g= 7.97 - 4.45 m 1 - 2.85 m 2+1.82 m 3 -3.38 m 4 -3.18 m 5 +1.13 m 6 +6.58 m 7 +4.67 m 8 +1.52 m 9 -7.73 m 10

[0043] Wherein, g represents the miRNA expression index, m 1 , m 2 , m 3 , m 4 , m 5 , m 6 , m 7 , m 8 , m 9 , m 10 respectively represent the expression levels of miR-31-5p, miR-219-5p, miR-1-3p, miR-130a-5p, miR-452-3p, miR-598-3p, miR-9-5p, miR-1246, miR-196a-5p and miR-210-3p.

[0044] V. A prognostic prediction system applied to the above prognostic prediction method

[0045] The prognostic prediction system includes:

[0046] An input module, configured to receive the clinical information corresponding to the in-vitro biological plasma sample to be predicted and the expression levels of each plasma miRNA; the clinical variables in the clinical information include age, KPS, and GPA;

[0047] A prognostic model module, configured to receive the clinical information and the expression levels of all plasma miRNAs through the prognostic model, obtain the predicted survival rate and output it;

[0048] The prognostic model is specifically:

[0049] h(t)=h 0 (t) exp( PI )

[0050] PI = 1.43 X 1 -0.28 X 2 -22.27 X 3 +58.72 X 4

[0051] wherein, h(t) represents the risk function, h 0 (t) represents the baseline risk function, X 1 represents age, X 2 represents KPS, X 3 represents GPA, X 4 represents the miRNA expression index, t represents the predicted survival time, PI represents the prognostic index;

[0052] An output module, configured to output the predicted survival rate to the prognostic prediction system or visualize the predicted survival rate.

[0053] Optionally, the visualization of the predicted survival rate is implemented through a nomogram.

[0054] The beneficial effects of the present invention are as follows:

[0055] 1. The present invention provides a set of plasma miRNAs, which are closely related to the prognostic effect of lung cancer brain metastasis and can be used as prognostic markers for lung cancer brain metastasis for prognostic prediction of lung cancer brain metastasis.

[0056] 2. The present invention provides a method for constructing a prognostic model based on the expression levels of plasma miRNAs and clinical information using a regression analysis method. The prognostic model constructed using this method can be applied to the prognostic prediction of lung cancer brain metastasis, and both the prediction model and the prognostic prediction method have high sensitivity and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1Schematic diagram of mining candidate miRNA combinations through a public database in the embodiments of the present invention; wherein, A is a principal component analysis diagram, B is a volcano diagram, C is a schematic diagram of the intersection extraction process, D is a differential ranking diagram, E is a schematic diagram of the correlation between low expression of hsa-miR-1246 and low survival rate, and F is a schematic diagram of the correlation between high expression of hsa-miR-31-5p and low survival rate.

[0058] Figure 2 Schematic diagram of gene enrichment and pathway analysis in the embodiments of the present invention; wherein, A is a pathway analysis result diagram, B is a gene enrichment analysis result diagram, C is a z-value analysis result diagram, and D is a Hub gene analysis result diagram.

[0059] Figure 3 Schematic diagram of differential analysis of in vitro biological plasma samples in the embodiments of the present invention; A is the plasma miRNA expression levels in two groups of samples, B is the Spearman pairwise correlation analysis result diagram of plasma miRNA combinations, and C is the Chord chord diagram of plasma miRNA combinations.

[0060] Figure 4 Schematic diagram of the prognostic model obtained in the comparative example of the present invention; A is a nomogram, B is a calibration curve diagram, and C is a receiver operating characteristic curve diagram.

[0061] Figure 5 Schematic diagram of the prognostic model obtained in the embodiments of the present invention; A is a nomogram, B is a calibration curve diagram, and C is a receiver operating characteristic curve diagram.

[0062] Figure 6 Technical roadmap adopted in the embodiments of the present invention. Detailed implementation manners

[0063] The following further illustrates the present application in combination with embodiments, but is not limited thereto. Materials, reagents, etc. used in the following embodiments can be obtained from commercial channels without special instructions. In the present invention, for the convenience of description or in line with professional habits, some terms in certain sentences are abbreviated or simplified in English, and the corresponding relationships are as follows and will not be repeated hereinafter:

[0064] BrM, Brain Metastasis, brain metastasis;

[0065] miRNA, microRNA, microRNA;

[0066] AUC, Area Under Curve, area under the curve;

[0067] ROC, Receiver Operator Characteristic Curve, receiver operating characteristic curve;

[0068] NSCLC, Non-Small Lung Cancer Cell, non-small cell lung cancer

[0069] LUAD, Lung Adenocarcinoma, lung adenocarcinoma;

[0070] KPS, Karnofsky Performance Status, Karnofsky performance status;

[0071] GPA, Graded Prognostic Assessment, graded prognostic assessment;

[0072] RPM, Reads Per Million Mapped Reads, reads per million mapped reads;

[0073] z-score, z-score;

[0074] TCGA, The Cancer Genome Atlas, The Cancer Genome Atlas;

[0075] GEO, Gene Expression Omnibus, Gene Expression Omnibus;

[0076] ArrayExpress, ArrayExpress - Functional Genomics Data, ArrayExpress - Functional Genomics Data;

[0077] PI, Prognostic Index, prognostic index;

[0078] PCA, Principal Component Analysis, principal component analysis.

[0079] The specific embodiments of the present invention are as follows:

[0080] Embodiment

[0081] As Figure 6 shown, in this embodiment, a prognostic model for lung cancer brain metastasis was constructed through the following steps:

[0082] 1) In this embodiment, by comprehensively analyzing the miRNA sequencing data downloaded from three bioinformatics databases, a candidate miRNA combination was obtained. The results of the comprehensive analysis are as Figure 1 shown.

[0083] The comprehensive analysis was specifically achieved through the following process:

[0084] 1.1) Download miRNA expression data and clinical information of different samples from TCGA database, GEO database and ArrayExpress database respectively, merge the miRNA expression data from TCGA database, GEO database and ArrayExpress database, construct an expression matrix with miRNA numbers, counts, and RPM as column vectors, normalize the expression data in the expression matrix to obtain a normalized matrix, and calculate the expression z-value of each miRNA in different samples according to the normalized matrix;

[0085] Among them, the expression z-value of any miRNA in sample A is obtained through the following formula:

[0086] z-score miRNA =(counts / RPM sample -counts / RPM average ) / (counts / RPM SD )

[0087] In the formula, z-score miRNA represents the expression z-value of miRNA in sample A, counts / RPM sample represents the expression of miRNA in sample A, counts / RPM average represents the average expression of miRNA in all samples, counts / RPM SD represents the standard deviation of miRNA in all samples.

[0088] 1.2) Combine the group labels (lung cancer brain metastasis group and control group) in the clinical information, and perform principal component analysis on the expression of all miRNAs through the "FactoMineR" package and "factoextra" package of R language to obtain a combination of differentially expressed miRNAs between two groups composed of multiple differentially expressed miRNAs between two groups.

[0089] In this embodiment, the "edgeR" and "limma" packages are also used to analyze the combination of differentially expressed miRNAs between two groups, and the volcano plot is drawn through the "ggplot2" package to visualize the data analysis results to present the statistical significance and fold change of each differentially expressed miRNA between two groups. See A and B in Figure 1 for the principal component analysis result diagram and the volcano plot respectively.

[0090] 1.3) Through the "DESeq2" package, differentially rank each differentially expressed miRNA between two groups according to statistical significance and fold change (see the results in Figure 1D), from the miRNA combination with differential expression between the two groups, select the genes with differential expression in all three databases ( Figure 1 C), 73 miRNAs with differential expression between the lung cancer brain metastasis group and the control group were obtained in all three datasets, forming a common differentially expressed miRNA combination. Combining with the preset threshold, the top 60 miRNAs with larger differences and significant statistical significance were selected from the common differentially expressed miRNA combination as the significantly common differentially expressed miRNAs in all bioinformatics databases, forming a significantly common differentially expressed miRNA combination for use as candidate miRNAs.

[0091] In this embodiment, the preset fold change threshold and the preset statistical significance threshold are respectively set as:

[0092] |log 2 (FC)| > 2 and p.adj < 0.05

[0093] In the formula, |·| represents the absolute value of the parameter "·", FC represents the fold change, and p.adj represents the adjusted P value, that is, the statistical significance.

[0094] As Figure 1 shown in D, all 73 significantly common differentially expressed miRNAs are within the preset threshold range. Figure 1 Figure A is the principal component analysis diagram of miRNA expression data from the GEO database. Blue represents the control group (primary), and red represents the lung cancer brain metastasis group (metastasis). It can be seen that the control group and the lung cancer brain metastasis group can be preliminarily distinguished according to the miRNA expression data.

[0095] Figure 1 Figure B is the volcano plot of miRNA expression data from the TCGA LUAD database. Red represents the miRNAs upregulated in the lung cancer brain metastasis group samples compared with the control group samples; blue represents the miRNAs downregulated in the lung cancer brain metastasis group samples compared with the control group samples. It can be seen that there are many differentially expressed miRNAs between the control group and the lung cancer brain metastasis group. In this embodiment, the number of miRNAs from the TCGA database is 527. Compared with the control group samples, there are 37 miRNAs highly expressed (fold change < 1 / 2, P value < 0.05) and 31 miRNAs lowly expressed (fold change > 2, P value < 0.05) in the lung cancer brain metastasis group samples.

[0096] Figure 1 Figure D is the differential ranking diagram of miRNAs. As shown in the figure, the miRNA combination with differential expression between the two groups includes a total of 1292 miRNAs. Among the 60 miRNAs in the candidate miRNA combination, there are 27 miRNAs highly expressed and 33 miRNAs lowly expressed in the lung cancer brain metastasis group samples compared with the control group samples.

[0097] In this example, combining the clinical information of all samples, the Kaplan–Meier survival analysis method was used to perform survival analysis on each candidate miRNA in the candidate miRNA combination. For the survival analysis results, see Figure 1 E and F in Figure 1 As shown in E of Figure 1 , the low expression of hsa-miR-1246 was associated with a low survival rate (hazard ratio = 0.71, P value = 0.038 < 0.05). As shown in

[0098] F of Figure 2 .

[0099] Figure 2 A in Figure 2 shows the results of pathway analysis. It can be seen that the candidate miRNA combination has functions in multiple aspects such as the Wnt signaling pathway, cell migration, Ras signaling pathway, cell-cell junction, non-small cell lung cancer, and lung cancer brain metastasis. Figure 2 B in

[0100] Figure 2 shows the results of gene enrichment analysis. It can be seen that the GO functions are mainly enriched in functional pathways related to tumorigenesis and development such as the mitotic cell cycle and cell adhesion.

[0101] C in

[0100] Figure 2 is a z-value analysis result graph, which shows the z-values corresponding to each GO functional pathway calculated through the "GOplot" package. A positive z-value indicates positive regulation, and a negative z-value indicates negative regulation.

[0101] D in

[0101] is a Hub Genes analysis result graph, which further shows the associations between the candidate miRNAs and their respective associated target genes. It can be seen that the candidate miRNAs are closely related to many genes related to tumorigenesis and development such as Myc, GSK3β, and PRKCA.

[0101] 2) Quantitative analysis of miRNAs was performed on the in vitro biological plasma samples of lung cancer brain metastasis cases and healthy controls collected clinically, and the quantitative analysis results of each candidate miRNA were obtained. Through secondary screening, 10 plasma miRNAs were determined from the candidate miRNA combinations to form a plasma miRNA combination, which was used as a combined biomarker for prognostic prediction of lung cancer brain metastasis.

[0102] The 10 plasma miRNAs were miR-31-5p, miR-219-5p, miR-1-3p, miR-130a-5p, miR-452-3p, miR-598-3p, miR-9-5p, miR-1246, miR-196a-5p, and miR-210-3p, respectively.

[0103] The secondary screening was achieved through the following process:

[0104] 2.1) Collection of in vitro biological plasma samples: 2 - 3 mL of early morning fasting blood from the population with lung cancer brain metastasis or healthy controls was collected into an EDTA anticoagulant tube, gently shaken and mixed to prevent blood clotting, and centrifuged at 4°C and 4000 rpm for 10 min in a low-temperature centrifuge within 1 hour after blood collection. The pale yellow transparent upper plasma was taken and aliquoted into sterile and RNAase-free EP tubes, and immediately transferred to an ultra-low temperature freezer at -80°C for storage, obtaining in vitro biological plasma samples of lung cancer brain metastasis or normal in vitro biological plasma samples. They were transported to the laboratory for testing with dry ice.

[0105] In this example, the number of in vitro biological plasma samples of lung cancer brain metastasis and normal in vitro biological plasma samples was 25 each.

[0106] 2.2) Extraction of RNA from in vitro biological plasma samples: The in vitro biological plasma samples were taken out from the ultra-low temperature freezer at -80°C and immediately placed on ice. After they were completely dissolved, 200 μL of RNAase-free was taken and added to a centrifuge tube with a volume of 1.5 mL. 1 mL of Trizol lysate was added to the centrifuge tube on ice, and mixed thoroughly with a pipette. 1 / 5 volume of chloroform was added to the lysate, and shaken vigorously for 15 seconds until an emulsion was formed, and then left to stand at 4°C for 5 min. After centrifugation at 12000g and 4°C for 15 min, the solution was divided into three layers: a colorless aqueous phase (upper layer), a white intermediate layer, and a red organic layer.

[0107] Carefully aspirate the upper aqueous phase and transfer it to a new centrifuge tube. Then add an equal volume of isopropanol and mix by inverting the tube up and down. Let it stand at 4°C for 10 min. After centrifugation at 12000 g and 4°C for 10 min, a white precipitate can be seen. Resuspend the precipitate with 1 mL of 75% (V / V) ethanol (prepared with DEPC water), centrifuge at 12000 g and 4°C for 5 min, discard the supernatant, and air-dry in a clean environment. Add an appropriate amount of DEPC water to dissolve the precipitate, and measure the OD 260 / 280 ratio and concentration of the sample RNA solution using NanoDrop 2000.

[0108] 2.3) Reverse transcription: After measuring the concentration of the sample RNA solution, dilute the sample RNA solution to 450 ng / μL, prepare a genomic DNA removal system, and react at 42°C for 2 min to remove genomic DNA.

[0109] The composition of the genomic DNA removal system is shown in Table 1.

[0110] Table 1. Genomic DNA removal system

[0111] Component Volume <![CDATA[RNase-free ddH 2 O]]> 6 μL 5 * gDNA Wiper Mix 2 μL Total RNA extract 2 μL (450 ng / μL)

[0112] Subsequently, use stem-loop for miRNA reverse transcription, prepare a miRNA reverse transcription system, and the amplification conditions are 25°C for 5 min, 50°C for 15 min, and 85°C for 5 min in sequence.

[0113] The composition of the miRNA reverse transcription system is shown in Table 2.

[0114] Table 2. miRNA reverse transcription system

[0115] Component Volume <![CDATA[RNase-free ddH 2 O]]> 5 μL Previous step mixture 10 μL Stem-loop primer (2 μM) 1 μL 10*RT Mix 2 μL HiScript II Enzyme Mix 2 μL

[0116] 2.4) Real-time fluorescence quantitative PCR: Prepare a qPCR reaction system, and the amplification conditions are: pre-denaturation at 95°C for 3 min, 40 cycles of cycling reaction, each cycle including 95°C for 10 s and 60°C for 30 s, and melting curve (95°C for 15 s, 60°C for 60 s, and 95°C for 15 s). Use U6 as the miRNA internal reference, and use the relative quantification method to calculate the relative expression level of the target gene to obtain the relative expression level of each candidate miRNA.

[0117] The composition of the qPCR reaction system is shown in Table 3.

[0118] Table 3. qPCR reaction system

[0119] Component Volume 2*qPCR Master Mix 10 μL F-Primer (10 μM) 0.4 μL R-Primer (10 μM) 0.4 μL Template DNA 2 μL <![CDATA[ddH 2 O]]> 7.2 μL

[0120] The primer sequences used in the qPCR reaction system are shown as SEQ ID NO.1 to SEQ ID NO.21 in Table 4 respectively.

[0121] Table 4. Primer sequences used in the qPCR reaction system

[0122] ID Primer name Sequence (5’->3’) SEQ ID NO.1 miR-1246-F GCGGCAATGGATTTTTGG SEQ ID NO.2 miR-1246-RT GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACCCTGCT SEQ ID NO.3 miR-219-5p-F CGGCTGATTGTCCAAACGC SEQ ID NO.4 miR-219-5p-RT GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACAGAATT SEQ ID NO.5 miR-598-3p-F CGGCTACGTCATCGTTGTCA SEQ ID NO.6 miR-598-3p-RT GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACTGACGA SEQ ID NO.7 miR-9-5p-F GCGGCTCTTTGGTTATCTAGCT SEQ ID NO.8 miR-9-5p-RT GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACTCATAC SEQ ID NO.9 miR-31-5p-F GCAGGCAAGATGCTGGCA SEQ ID NO.10 miR-31-5p-RT GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACCAGCTA SEQ ID NO.11 miR-210-3p-F GGCCTGTGCGTGTGACAGC SEQ ID NO.12 miR-210-3p-RT GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACTCAGCC SEQ ID NO.13 miR-196a-5p-F CGCGGCTAGGTAGTTTCATGTT SEQ ID NO.14 miR-196a-5p-RT GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACCCCAAC SEQ ID NO.15 miR-130a-5p-F GCGGCTTCACATTGTGCTACT SEQ ID NO.16 miR-130a-5p-RT GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACGCAGAC SEQ ID NO.17 miR-1-3p-F CCGGCTGGAATGTAAAGAAGT SEQ ID NO.18 miR-1-3p-RT GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACATACAT SEQ ID NO.19 Common-R CAGTGCGTGTCGTGGAGT SEQ ID NO.20 U6-F CTCGCTTCGGCAGCACA SEQ ID NO.21 U6-R AACGCTTCACGAATTTGCGT

[0123] 2.5) Construct a sampling data set with the relative expression levels of candidate miRNAs in all ex vivo biological plasma samples, perform Spearman pairwise correlation analysis on the sampling data set and eliminate collinear miRNAs, and then perform differential expression analysis to obtain 10 plasma miRNAs with differential expression between the two groups, forming a plasma miRNA combination, which is used as a combined biomarker for the prognosis prediction of lung cancer brain metastasis.

[0124] The plasma miRNAs are miR-31-5p, miR-219-5p, miR-1-3p, miR-130a-5p, miR-452-3p, miR-598-3p, miR-9-5p, miR-1246, miR-196a-5p and miR-210-3p respectively.

[0125] As Figure 3 shown in A, the expression levels of miR-31-5p, miR-219-5p, miR-1-3p, miR-130a-5p and miR-452-3p were significantly up-regulated in ex vivo biological plasma samples of lung cancer brain metastasis (P < 0.001), and the expression levels of miR-598-3p, miR-9-5p, miR-1246, miR-196a-5p and miR-210-3p were significantly down-regulated in ex vivo biological plasma samples of lung cancer brain metastasis (P < 0.001).

[0126] Perform Spearman pairwise correlation analysis on the plasma miRNA combination again to further verify the correlation between plasma miRNAs. As Figure 3 shown in B, from the results of Spearman pairwise correlation analysis, it can be seen that the plasma miRNA combination is divided into two subgroups. The first subgroup consists of miR-31-5p, miR-219-5p, miR-1-3p, miR-130a-5p and miR-452-3p, and the second subgroup consists of miR-598-3p, miR-9-5p, miR-1246, miR-196a-5p and miR-210-3p. The miRNAs within each subgroup are positively correlated, and the two subgroups are negatively correlated (P < 0.05). This result is consistent with Figure 3 the result in A, further corroborating the stability and reliability of plasma miRNAs as biomarkers. AsFigure 3 As shown in C of, the Chord diagram further shows the pairwise correlations of 10 plasma miRNAs. The width and depth of the bands between the color blocks reflect the correlations between the two.

[0127] 2.6) Using the Logistic regression method, a prediction model of miRNA expression index was constructed based on the plasma miRNA combination and the expression levels of individual plasma miRNAs.

[0128] The prediction model of miRNA expression index is specifically as follows:

[0129] g= 7.97 - 4.45 m 1 -2.85 m 2 +1.82 m 3 -3.38 m 4 -3.18 m 5 +1.13 m 6 +6.58 m 7 +4.67 m 8 +1.52 m 9 -7.73 m 10

[0130] In the formula, g represents the miRNA expression index, m 1 、 m 2 、 m 3 、 m 4 、 m 5 、 m 6 、 m 7 、 m 8 、 m 9 、 m 10 respectively represent the expression levels of miR-31-5p, miR-219-5p, miR-1-3p, miR-130a-5p, miR-452-3p, miR-598-3p, miR-9-5p, miR-1246, miR-196a-5p, and miR-210-3p.

[0131] 3) Compose the clinical information corresponding to each in vitro biological plasma sample of lung cancer brain metastasis collected clinically into a regression analysis data set, perform univariate Cox regression analysis and multivariate Cox regression analysis on the regression analysis data set in sequence. According to the analysis results, select age, KPS, and GPA that are significantly correlated with prognosis (P < 0.05). These clinical variables form a clinical variable combination. Use the Cox regression analysis method to construct a Cox regression prognosis prediction model based on the clinical variable combination and miRNA expression index, and use the K-fold method to optimize the Cox regression prognosis prediction model to obtain a prognosis model.

[0132] In this embodiment, the clinical information includes age, gender, smoking history, tumor TNM stage, pathological type, history of radiotherapy and chemotherapy, history of targeted drug use, KPS, tumor metastasis focus, detection of EGFR, PD-L1, KRAS, and ALK gene mutation sites, and survival status.

[0133] The prognosis model obtained in this embodiment is specifically:

[0134] h(t)=h 0 (t) exp( PI )

[0135] PI = 1.43 X 1 -0.28 X 2 -22.27 X 3 +58.72 X 4

[0136] In the formula, h(t) represents the risk function, h 0 (t) represents the baseline risk function, X 1 represents age, X 2 represents KPS, X 3 represents GPA, X 4 represents the miRNA expression index, t represents the predicted survival time, PI represents the prognosis index.

[0137] 4) After obtaining the prognosis model, use the "survival" package and "nomogramFormula" package in R language to draw a nomogram to visualize the prognosis model. The nomogram of the prognosis model obtained in this embodiment is shown inFigure 5 A, by calculating the prognostic index value PI can predict the predicted survival time t within the predicted survival rate S(t) .

[0138] S(t) = exp(- ∫ t 0 h(u)du ) = exp(- ∫ t 0 h 0 (u)du × exp( PI ))

[0139] In the formula, S(t) is the survival function, that is, the probability that the individual survival time T is greater than the predicted survival time t , P(T>t) , t is the predicted survival time, with the unit of year, ∫ t 0 h 0 (u)du represents the cumulative risk within the predicted survival time t .

[0140] In this embodiment, through the above process, the one-year survival rate S(1)。 Figure 5 B of this embodiment is the calibration curve graph of the prognostic model of this embodiment. The abscissa in the graph is the survival probability predicted by the model, and the ordinate is the actually observed survival probability. Each point represents the situation of the survival probability predicted by the model and the actually observed survival probability. The gray diagonal line is the line in the ideal situation. As shown in the figure, the 1-year survival rate (blue line) has a high degree of fitting with the diagonal line, and the prediction effect is good. After calculation, the concordance index of this model is 0.82.

[0141] Subsequently, the receiver operating characteristic curve is drawn through the "pROC" package of R language, and the area under the curve is calculated. The results are shown in Figure 5 C. The abscissa is 1-specificity, representing the false positive rate, and the ordinate is the sensitivity, representing the true positive rate. The point closest to the upper left of the coordinate axis in this graph is the critical value with relatively high sensitivity and specificity. The red represents the curve of the prognostic model obtained in this embodiment, and the blue represents the curve of the prognostic model obtained in the comparative example. The area under the curve of the prognostic model in this embodiment is 0.838.

[0142] Comparative example

[0143] This comparative example is based on the clinical variable combination (age, KPS, and GPA) obtained in step 3) of the example. The Cox regression analysis method is used to construct a Cox regression prognosis prediction model, and the K-fold method is used to optimize the Cox regression prognosis prediction model to obtain a prognosis model.

[0144] The prognosis model obtained in this comparative example is specifically:

[0145] h(t)=h 0 (t) exp(1.50 X 1 -0.30 X 2 -6.09 X 3 )

[0146] PI =1.50 X 1 -0.30 X 2 -6.09 X 3

[0147] In the formula, h(t) represents the risk function, h 0 (t) represents the baseline risk function, X 1 represents age, X 2 represents KPS, X 3 represents GPA, t represents the survival time, PI represents the prognostic index.

[0148] After obtaining the prognosis model, the "survival" package and the "nomogramFormula" package in R language are used to draw a nomogram to visualize the prognosis model. The nomogram of the prognosis model is shown in Figure 4 A. By calculating the prognostic index value PI to predict the 1-year survival rate S(1) .

[0149] S(t) =exp(- ∫ t 0 h(u)du ) = exp(- ∫ t 0 h 0(u)du × exp( PI ))

[0150] In the formula, S(t) represents the survival function, that is, the probability that the individual survival time T is greater than the predicted survival time t ; P(T>t) , t represents the predicted survival time, with the unit of year, ∫ t 0 h 0 (u)du represents the predicted survival time t within the cumulative risk.

[0151] Figure 4 Figure B of this comparative example prognosis model is the calibration curve graph. In the graph, the abscissa is the survival probability predicted by the model, and the ordinate is the actually observed survival probability. Each point represents the situation of the survival probability predicted by the model and the actually observed survival probability. The gray diagonal line is the line of the ideal situation. As shown in the figure, the 1-year survival rate (blue line) has a certain degree of fitting with the diagonal line, indicating that the model has a certain prognostic prediction effect on the 1-year survival rate. The concordance index (C-index) of this model is calculated to be 0.67.

[0152] Subsequently, the receiver operating characteristic curve was drawn through the "pROC" package in R language, and the area under the curve was calculated. Figure 4 Figure C of this invention's comparative example prognosis model is the receiver operating characteristic curve. The abscissa is 1 - specificity, representing the false positive rate, and the ordinate is the sensitivity, representing the true positive rate. The point closest to the upper left of the coordinate axis in this graph is the critical value with relatively high sensitivity and specificity. After calculation, the area under the curve is 0.669 (P < 0.01).

[0153] It can be seen that after adding the miRNA expression index, the AUC value of the prognosis model is significantly improved (P < 0.01), and the prediction effect is significantly improved. Using the plasma miRNA combination obtained in this invention as a combined biomarker and applying it to construct a prognosis model for lung cancer brain metastasis can effectively improve the accuracy of the prognosis model. The prognosis model obtained by the construction method of this invention for the prognosis prediction of lung cancer brain metastasis also shows a higher accuracy.

[0154] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

[0155] The nucleotide sequences involved in the present invention are specifically as follows:

[0156] SEQ ID NO.1: Name: Nucleotide sequence of miR-1246-F; DNA type: DNA (other DNA); Organism source: synthetic construct; GCGGCAATGGATTTTTGG;

[0157] SEQ ID NO.2: Name: Nucleotide sequence of miR-1246-RT; DNA type: DNA (other DNA); Organism source: synthetic construct; GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACCCTGCT;

[0158] SEQ ID NO.3: Name: Nucleotide sequence of miR-219-5p-F; DNA type: DNA (other DNA); Organism source: synthetic construct; CGGCTGATTGTCCAAACGC;

[0159] SEQ ID NO.4: Name: Nucleotide sequence of miR-219-5p-RT; DNA type: DNA (other DNA); Organism source: synthetic construct; GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACAGAATT;

[0160] SEQ ID NO.5: Name: Nucleotide sequence of miR-598-3p-F; DNA type: DNA (other DNA); Organism source: synthetic construct; CGGCTACGTCATCGTTGTCA;

[0161] SEQ ID NO.6: Name: Nucleotide sequence of miR-598-3p-RT; DNA type: DNA (other DNA); Organism source: synthetic construct; GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACTGACGA;

[0162] SEQ ID NO.7: Name: Nucleotide sequence of miR-9-5p-F; DNA type: DNA (other DNA); Organism source: synthetic construct; GCGGCTCTTTGGTTATCTAGCT;

[0163] SEQ ID NO.8: Name: Nucleotide sequence of miR-9-5p-RT; DNA type: DNA (other DNA); Organism source: synthetic construct; GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACTCATAC;

[0164] SEQ ID NO.9: Name: Nucleotide sequence of miR-31-5p-F; DNA type: DNA (other DNA); Organism source: synthetic construct; GCAGGCAAGATGCTGGCA;

[0165] SEQ ID NO.10: Name: Nucleotide sequence of miR-31-5p-RT; DNA type: DNA (other DNA); Organism source: synthetic construct; GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACCAGCTA;

[0166] SEQ ID NO.11: Name: Nucleotide sequence of miR-210-3p-F; DNA type: DNA (other DNA); Organism source: synthetic construct; GGCCTGTGCGTGTGACAGC;

[0167] SEQ ID NO.12: Name: Nucleotide sequence of miR-210-3p-RT; DNA type: DNA (other DNA); Organism source: synthetic construct; GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACTCAGCC;

[0168] SEQ ID NO.13: Name: Nucleotide sequence of miR-196a-5p-F; DNA type: DNA (other DNA); Organism source: synthetic construct; CGCGGCTAGGTAGTTTCATGTT;

[0169] SEQ ID NO.14: Name: Nucleotide sequence of miR-196a-5p-RT; DNA type: DNA (other DNA); Organism source: synthetic construct; GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACCCCAAC;

[0170] SEQ ID NO.15: Name: Nucleotide sequence of miR-130a-5p-F; DNA type: DNA (other DNA); Organism source: synthetic construct; GCGGCTTCACATTGTGCTACT;

[0171] SEQ ID NO.16: Name: Nucleotide sequence of miR-130a-5p-RT; DNA type: DNA (other DNA); Organism source: synthetic construct; GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACGCAGAC;

[0172] SEQ ID NO.17: Name: Nucleotide sequence of miR-1-3p-F; DNA type: DNA (other DNA); Organism source: synthetic construct; CCGGCTGGAATGTAAAGAAGT;

[0173] SEQ ID NO.18: Name: Nucleotide sequence of miR-1-3p-RT; DNA type: DNA (other DNA); Organism source: synthetic construct; GTCGTATCCAGTGCGTGTCGTGGAGTCGGCAATTGCACTGGATACGACATACAT;

[0174] SEQ ID NO.19: Name: Nucleotide sequence of Common-R; DNA type: DNA (other DNA); Organism source: synthetic construct; CAGTGCGTGTCGTGGAGT;

[0175] SEQ ID NO.20: Name: Nucleotide sequence of U6-F; DNA type: DNA (other DNA); Organism source: synthetic construct; CTCGCTTCGGCAGCACA;

[0176] SEQ ID NO.21: Name: Nucleotide sequence of U6-R; DNA type: DNA (other DNA); Organism source: synthetic construct; AACGCTTCACGAATTTGCGT.

Claims

1. A plasma miRNA combination marker for predicting the prognosis of brain metastasis of lung cancer, characterized in that: The plasma miRNA combination marker is mainly composed of miR-31-5p, miR-219-5p, miR-1-3p, miR-130a-5p, miR-452-3p, miR-598-3p, miR-9-5p, miR-1246, miR-196a-5p and miR-210-3p.

2. The plasma miRNA combination marker according to claim 1, characterized in that: The plasma miRNA is the miRNA in an in vitro biological plasma sample.

3. Use of a reagent for detecting the plasma miRNA combination marker according to claim 1 or 2, characterized in that: Used for preparing a prognosis prediction drug or a prognosis prediction system for lung cancer brain metastasis, wherein the prognosis prediction drug comprises a primer, a probe or a fluorescent reaction agent capable of specifically detecting all plasma miRNAs.

4. A prognosis prediction system for lung cancer brain metastasis, applied to the prognosis prediction method for lung cancer brain metastasis based on the plasma miRNA combination marker as claimed in claim 1 or 2, characterized in that: The prognosis prediction system comprises: An input module, used for receiving clinical information corresponding to the in vitro biological plasma sample to be predicted and the expression level of each plasma miRNA; The prognostic model module is used to input clinical information and the expression levels of all plasma miRNAs into the prognostic model to obtain the predicted survival rate; An output module, used to output the predicted survival rate to the prognosis prediction system or to visualize the predicted survival rate; The prognosis prediction method comprises the following steps: obtaining the expression level of each plasma miRNA in an in vitro biological plasma sample and the clinical information corresponding to the in vitro biological plasma sample, inputting the expression levels of all plasma miRNAs and the clinical information into a prognosis model, the prognosis model obtains a miRNA expression index according to the expression levels of all plasma miRNAs, and then obtains a predicted survival rate according to the miRNA expression index and the clinical information and outputs it as a prognosis prediction result.

5. The prognosis prediction system according to claim 4, characterized in that: The clinical variables in the clinical information include age, KPS and GPA, and the predicted survival rate is obtained by the following formula: S(t) =exp(- ∫ t 0 h 0 (u)du ×exp( PI )) PI =1.43 X 1 -0.28 X 2 -22.27 X 3 +58.72 X 4 In the formula, S(t) Represents the predicted survival time t The predicted survival rate, t represents the predicted survival time, ∫ t 0 h 0 (u)du Represents the predicted survival time t The cumulative risk within PI represents the prognostic index, X 1 Indicates age, X 2 represents KPS, X 3 Indicates GPA, X 4 Represents the miRNA expression index.

6. The prognosis prediction system according to claim 5, characterized in that: The miRNA expression index is obtained by the following formula: g= 7.97-4.45 m 1 -2.85 m 2 +1.82 m 3 -3.38 m 4 -3.18 m 5 +1.13 m 6 +6.58 m 7 +4.67 m 8 +1.52 m 9 -7.73 m 10 In the formula, g represents the miRNA expression index, m 1 , m 2 , m 3 , m 4 , m 5 , m 6 , m 7 , m 8 , m 9 , m 10 They represent the expression levels of miR-31-5p, miR-219-5p, miR-1-3p, miR-130a-5p, miR-452-3p, miR-598-3p, miR-9-5p, miR-1246, miR-196a-5p and miR-210-3p, respectively.

Images